- Email: [email protected]
Refrigerant distribution in the vertical header of the microchannel heat exchanger – Measurement and visualization of R410A flow
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Available online at www.sciencedirect.com
w w w . i i fi i r . o r g
journal homepage: www.elsevier.com/locate/ijrefrig
Refrigerant distribution in the vertical header of the microchannel heat exchanger e Measurement and visualization of R410A flow Yang Zou a, Pega S. Hrnjak a,b,* a
Air Conditioning and Refrigeration Center, Department of Mechanical Science and Engineering, University of Illinois at Urbana Champaign, 1206 West Green Street, Urbana, IL 61801, United States b Creative Thermal Solutions Inc., 2209 North Willow Road, Urbana, IL 61802, United States
article info
abstract
Article history:
This paper presents the R410A adiabatic upward flow in three vertical headers of micro-
Received 26 November 2012
channel heat exchangers. All microchannel tubes are inserted into the half depth. The ob-
Received in revised form
jectives are to explore what affects R410A distribution and attempt to predict the distribution.
30 March 2013
R410A is circulated into the header through the (5 or 10) tubes in the bottom pass and exits
Accepted 30 April 2013
through the (5 or 10) tubes in the top pass. It represents the flow in the outdoor heat exchanger
Available online 9 May 2013
(usually used as the condenser) when it is used as the evaporator in the heat pump mode of reversible systems. The quality was typically varied from 0.2 to 0.8 and the mass flow rate
Keywords:
from 1.5 to 4.5 kg h1 per tube. The best distribution was observed at high mass flux and low
Two-phase flow
quality. The experiment and visualization reveals the flow pattern effects in terms of ho-
Refrigerant distribution
mogeneity and momentum. The churn flow had better distribution since the two-phase
Vertical header
mixture was more homogenous and the distribution was better at high mass flux in the
Microchannel
header because the higher momentum liquid was able to reach the top exit tube.
Heat exchanger
ª 2013 Elsevier Ltd and IIR. All rights reserved.
Distribution du frigorige`ne dans le collecteur vertical d’un e´changeur de chaleur a` microcanaux e Mesures et visualisation de l’e´coulement du R410A Mots cle´s : e´coulement diphasique ; distribution du frigorige`ne ; collecteur vertical ; microcanal ; e´changeur de chaleur
1.
Introduction
The microchannel heat exchangers have come to the frontier of stationary air conditioning and refrigeration applications
after its success in the mobile air conditioning systems, for their advantages in compactness and higher air-side heat transfer coefficients. In the reversible systems, the microchannel heat exchanger functions as both the evaporator and
* Corresponding author. Air Conditioning and Refrigeration Center, Department of Mechanical Science and Engineering, University of Illinois at Urbana Champaign, 1206 West Green Street, Urbana, IL 61801, United States. Tel.: þ1 217 244 6377; fax: þ1 217 244 6534. E-mail address: [email protected] (P.S. Hrnjak). 0140-7007/$ e see front matter ª 2013 Elsevier Ltd and IIR. All rights reserved. http://dx.doi.org/10.1016/j.ijrefrig.2013.04.021
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Nomenclature A D CF G H h i j L LF m n Q P Re T U x y
2
Cross section area in the header (m ) Internal diameter of the header (m) Correction factor () Mass flux (kg m2 s1) Liquid reach (m) Liquid separation height (m) Enthalpy (kJ kg1) Superficial velocity (m s1) Length of the top exit part in header (m) Liquid fraction () Mass flow rate (g s-1) number of the outlet tubes () Power of heaters (kW) Pressure (kPa) Reynolds number () Temperature (K) A function Quality () Parameter
condenser. The outdoor heat exchanger, typically a multipass microchannel heat exchanger, has vertical headers to reduce the cost and allow for easier bending when a “U-shaped” condenser is needed. In the cooling mode, the outdoor heat exchanger functions as the condenser and the overall flow is downward. In the heating mode, it functions as the evaporator and the refrigerant flows in the reverse direction. Refrigerant maldistribution due to the two-phase flow in the header deteriorates the heat exchanger’s performance, and consequently reduces the system efficiency. Kulkarni et al. (2004) simulated the effect of R410A maldistribution induced by the pressure drop in the horizontal header. The microchannel evaporator’s capacity was reduced by 20%. Brix et al. (2009, 2010) modeled R134a and R744 distribution in two microchannel tubes, respectively. Due to maldistribution, the capacity was reduced by 23% and 18%, respectively. Good designs that provide reasonable distribution in either the air conditioning or the heat pump modes are still a challenge, although it has been extensively studied. Hrnjak (2004) and Webb and Chung (2005) have reviewed the refrigerant distribution issues in the header. Combined with others’ findings, it can be conceived that maldistribution is a very complex problem that is affected by numerous parameters: header geometry and orientation, fluid properties and inlet conditions, etc. Specifically, in their studies of horizontal headers, Fei and Hrnjak (2004), Vist and Pettersen (2004), Bowers et al. (2006) and Hwang et al. (2007) showed that the flow regimes in the header, which is affected by several conditions, had a strong influence on liquid distribution into the branch tubes in two-phase flow. In the studies of vertical headers, Moura (1990) examined airewater distribution in the second-pass header. In either the upward or downward flow in the header, phase separation occurred. The flow patterns strongly affected the distribution. Watanabe et al. (1995) investigated the distribution of R11 in a header with five round branch tubes. Along the upward flow, the flow regime transited from annular to slug, or froth flow, in the header.
G m r s
2197
Liquid take-off ratio () Viscosity (kg m1 s1) Density (kg m3) Coefficient of variation ()
Superscripts c Corrected meas Measured Subscripts ave Average pressure i Branch number in In the middle of the header and at the smallest area of the last inlet tube l Liquid M Main pipe (header) out Out from the header sup Superheated point sub Subcooled point v vapor
Increasing the quality to 0.3 improved the distribution because the higher velocity enabled liquid to reach the top tube. Song et al. (2002) manifested CO2 maldistribution in a multi-pass outdoor microchannel coil during the heating mode through frosting patterns. The balance between inertial, gravitational and shear forces generated two different distributions: 1) the liquid was only present in the bottom tubes at low qualities; 2) the top and bottom tubes received less liquid than the middle tubes at high qualities. Cho and Cho (2004) tested R22 distribution in the header with the microchannel tubes. Most liquid was in the bottom due to gravity regardless of the inlet types and inlet quality. Through infrared images, Dschida and Hrnjak (2008) showed R410A maldistribution in a multi-pass microchannel heat exchanger during the heat pump mode. The superheated zone always appeared at the top in each pass, i.e. less liquid at the top, if any. The visualization of upward air and water flow in the header from Lee (2009) showed that three distinct regions were formed with different flow patterns, which caused three different distribution patterns. Byun and Kim (2011) tested and visualized R410A distribution in both the inlet and second-pass headers at one inlet quality condition. For the inlet header, a liquid pool was formed at the bottom. For the second pass header, an upward-flow liquid film was formed in the header. Because of the high axial momentum, most liquid was taken by the middle tubes. Increasing the mass flux resulted in most liquid in the top tubes. Although this study is similar to Byun and Kim (2011), this study uses a different experiment method so that the mass flow rate in each tube is measured and this study covers a much wider range of inlet mass flux and quality. The effects of tubes number is also explored. To predict the refrigerant distribution, either numerical simulation or empirical correlation was developed. Moura (1990) and Tompkins et al. (2002) developed their models based on finite volume method. Fei and Hrnjak (2004) conducted CFD simulation with the commercial software FLUENT 6. Lee (2009) considered the vertical header as a series of T-
2198
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
junctions, and predicted the distribution based on the studies of two-phase flow at T-junction. Watanabe et al. (1995) and Byun and Kim (2011) developed empirical correlations based on the defined liquid and vapor take-off ratio. It is the ratio of liquid (or vapor) mass flow rate in the branch tube to the liquid (or vapor) mass flow rate in the main pipe (header) at immediate upstream, which is the region in the header right before the two-phase fluid exits through the branch tube. The liquid (or vapor) take-off ratio is related to the vapor phase Reynolds number in the main pipe at immediate upstream. This method will be used to develop an empirical correlation in this paper. It is noteworthy that most of the previous tests were done in the horizontal headers and low qualities (i.e. <0.5) because the indoor heat exchanger and/or inlet header were usually studied. However, typical outdoor microchannel heat exchangers have vertical headers and at least two passes. The two-phase refrigerant is redistributed in the intermediate header in which case the distribution phenomenon is more complex but important. The previous studies on multi-pass heat exchangers with nonintrusive methods, such as infrared imaging and frost accumulation, did not provide enough information to quantify refrigerant distribution. Thus, in this paper, R410A distribution in the vertical intermediate header of a multi-pass heat exchanger (with capacity of 0.5e2.5 kW) was studied by adding individual flow rate measurement and visualization.
2.
Experimental method
The test loop was constructed to study R410A distribution in the microchannel heat exchanger, as shown in Fig. 1. The liquid refrigerant was pumped into the inlet header while the inlet mass flow rate min was measured by the mass flow meter (Micromotion D40, 0.75%). The mass flow rate was controlled by a bypass valve and variable frequency drive. The temperature Tsub (immersed thermocouple, 0.5 C) and pressure Psub (Sensotec TJE, 1% FS) at the inlet to the first header were
measured to determine the subcooling. The subcooled liquid was assumed to distribute evenly into the microchannel tubes in the bottom pass, where the refrigerant was heated to the desired quality while the heaters are insulated. The set of six heaters per microchannel tube were providing uniform heat supply. The power needed to determine the inlet quality was measured by the watt transducer (Ohio Semitronics GW5024CX5, 0.2% FS). After the refrigerant entered into the lower part of the header, it turned 90 to flow upward and reach the upper part of the header. Due to maldistribution, different amounts of liquid exited through the microchannel tubes in the top pass. In these tubes, the refrigerant was heated again to provide equal superheat at the exit. Each tube was heated by six heaters and insulated. Each set of six heaters was individually controlled to provide adequate power to generate equal superheat, and was also individually measured by watt transducers (Ohio Semitronics GW5024CX5, 0.2% FS) to determine inlet quality into each exit tube. The temperature Tsup,i (immersed thermocouple, 0.5 C) in each tube and pressure Psup (Sensotec TJE, 1% FS) were measured to determine the superheat. The mass flow rate of the single-phase superheated vapor was measured individually as the total mass flow rate for each microchannel tube by the mass flow meter (Micromotion D06, 0.15%). The vapor was then brought to the condenser. With the help of the receiver and the subcooler, the subcooled liquid was provided to the pump. The thermal physical and thermodynamic properties are calculated in EES (2010). Enthalpy at the superheated point isup,i was determined by Tsup,i and Psup. The outlet enthalpy from the header, i.e. inlet to each exit tube, iout,i, was determined as iout;i ¼ isup;i
Q_ out;i _ out;i m
(1)
The pressure in the header Pave is estimated as the average of the subcooled and superheating pressures. The outlet quality from the header xout,i can be found from iout,i and Pave. Then, the liquid mass flow rate in each tube can be obtained as
Fig. 1 e System schematics.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
_ out;i 1 xout;i _ l;out;i ¼ m m
(2)
A high speed camera, Phantom v4.2, is used for visualizing the flow in the transparent header. The exposure time of the camera was set from 80 ms to 100 ms. The framing rate was at 2200 frames per second. The resolution was set as 512 512 or 256 512 pixels. Three circular headers were examined, each with all the microchannel tubes inserted to the half depth: 1) the aluminum header with five inlet and five exit microchannel tubes as used in industry (5 þ 5 aluminum header), 2) the transparent replica of the first (5 þ 5 transparent header), and 3) the transparent header with ten inlet and ten exit microchannel tubes (10 þ 10 transparent header) to study the effect of increasing microchannel number and header length. The transparent headers were made of PVC tube. The gap between the microchannel tube and PVC tube was sealed by a special epoxy adhesive. It is the black material on the right part of the header in the images. In addition, an epoxy block was made to increase the pressure tolerance while still ensuring the transparency outside the PVC tube. The measured inner diameter of the aluminum header is 14.94 mm, and that of the transparent header is 15.44 mm. The 5 þ 5 header is 170 mm long, and the 10 þ 10 header is 300 mm long. The tube pitch is 13 mm. The tubes are 1200 mm long. As shown in Fig. 1, each microchannel tube has 17 rectangular channels. Based on the manufacturer data, each microchannel is 0.5 mm 0.54 mm. The hydraulic diameter is calculated to be about 0.5 mm. To examine the case of ten inlet and ten exit tubes using the same header diameter (different circuiting), modification of the facility was made with intention to maintain the same number of mass flow meters (five) and the same heaters and control strategy. Two neighboring exit tubes were combined into one with a “Y” connection and the refrigerant was heated together to be superheated. Assuming that this change affected resolution of the distribution parameters only, the essence of the results should not change. Therefore, only five liquid fraction results were shown, whereas actually each one represented the liquid amount in the neighboring two branches. As shown in Fig. 1, in order to facilitate the connection of 10 þ 10 header, only five inlet microchannel tubes were connected and allowed the refrigerant to pass through implicitly assuming the insignificant effect of inlet difference. The other five inlet microchannel tubes were blocked and just served as obstructions in the header. The averaged saturation temperature was about 5 C in the header. The inlet quality ranged from 0.2 to 0.8. The results of xin ¼ 0 and 1 were taken to provide two asymptotes, and those of xin ¼ 0.95 were to offer more typical situations at high quality. The mass flow rate was from 1.5 to 4.5 kg h1 for each microchannel tube. Thus, the inlet mass flow rate ranged from 2.14 to 6.25 g s1 for the 5 þ 5 header and from 4.28 to 12.50 g s1 for the 10 þ 10 header. Tests at 7.27 g s1 for the 5 þ 5 header were taken in attempt to reach better distribution. Tests at 2.22 g s1 for the 10 þ 10 header were taken to compare the results with those at 2.14 g s1 for the 5 þ 5 header. From Bowers et al. (2006), the mass flux defined by the smallest cross sectional area in the header is more representative when the protrusion is present, so the maximum mass
2199
flux in the header Gin, as shown in Fig. 1, will be used in the following analysis. And Gin ranged from 21.80 to 63.67 kg m2 s1 for the 5 þ 5 headers and from 22.62 to 127.3 kg m1 s1 for the 10 þ 10 header.
3.
Data reduction
Two metrics are used to evaluate the distribution. The first one is coefficient of variation s. It is the dimensionless standard deviation, which is defined as s¼
1 ml
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi n 2 1X _ l;out;i m l m n 1
(3)
The uniform distribution is described by s ¼ 0. The worst distribution, for this case (with five results), is when s ¼ 2. The advantage of this metric is that one number is able to characterize the goodness of distribution. However, in order to illustrate the distribution profile, it is necessary to apply the liquid fraction, LFi ¼
_ l;out;i m n P _ l;out;i m
(4)
i
For the case with five results, the uniform distribution is described by LFi ¼ 0.2 in every tube. The uncertainty propagation analysis of s and LFi was carried out in the EES (2010). It was based on vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u N uX vU dyi where U ¼ U y1 ; y2 ; y3 ; ::: dU ¼ t vy i i¼1
(5)
Thus, the uncertainty of s was within 4.5%. The uncertainty of liquid fraction is usually below 5%. Due to the uncertainty of the instruments such as the mass flow meter and watt transducer, the sum of the mass flow rates in the outlet tubes is not the same as the inlet mass flow rate. To analyze the data and develop a correlation, the mass flow rate in the outlet tube is corrected as in equations (6) and (7). _ meas _ cout;i ¼ m m out;i CF
(6)
_ meas m CF ¼ Pn in meas _ i¼1 mout;i
(7)
In addition, the liquid mass flow rates in the outlet tubes are adjusted such that the sum is equal to the inlet liquid mass flow rate, as in equations (8) and (9). _ meas _ cl;out;i ¼ m m l;out;i CFl
CFl ¼
_ meas 1 xmeas m in in Pn _ meas i¼1 ml;out;i
(8)
(9)
The typical CF value is around 0.9. For the 10 þ 10 header, only five liquid flow rate results were obtained. To estimate the flow rate in each tube for further analysis, it is assumed that the mass flow rate (either liquid or vapor) of the neighboring two tubes is equal.
2200
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
4.
Results and discussion
4.1.
Distribution profiles
Figs. 2e4 show the distribution profiles and the coefficient of variation of the 5 þ 5 aluminum header (Fig. 2), 5 þ 5 transparent header (Fig. 3) and 10 þ 10 transparent header (Fig. 4). The darkness of the bar color represents different branches (tubes), the pale being the lowest exit branch and the dark being the highest exit branch. The best distribution indicated
by s is at xin ¼ 0.2 (the lowest) and the highest inlet mass flux among the test conditions for each header. The results of the 5 þ 5 aluminum and transparent headers are very similar (Figs. 2 and 3). The small discrepancy may be due to the different internal areas. In Figs. 2(a) and 3, the liquid fraction of the top branch is higher as the mass flow rate increases and at xin ¼ 0.6, 0.8 and 0.95, the liquid fractions of the bottom branches reduce at the same time, whereas at xin ¼ 0.2 and 0.4, the liquid fractions of the bottom branches do not change much. On the other hand, the liquid fraction of the highest branch reduces as the quality increases. Meanwhile,
Fig. 2 e Distribution results of the 5 D 5 aluminum header. a) Liquid fraction, b) Coefficient of variation.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
2201
Fig. 3 e Distribution results of the 5 D 5 transparent header.
for the other branches, as quality increases, 1) at min ¼ 2.14 and 3.17 g s1, the bottom branches have higher and higher liquid fractions; 2) at min ¼ 4.19 g s1, more liquid leaves from the middle tubes; 3) at min ¼ 5.22, 6.25 and 7.27 g s1, the liquid fraction of branch 4 is highest while the others have less and less. The values of the coefficient of variation in Fig. 2(b) show that: 1) at the fixed mass flow rate, the distribution usually
deteriorates as quality increases; 2) at low qualities, the distribution improves with increasing mass flux; 3) at high qualities, the distribution is better at some intermediate mass flow rate. It gets worse as min either increases or decreases. It is because as min increases, the bottom tubes liquid rich transited to the top tubes liquid rich. During this transition, there is a case that neither the bottom tubes nor the top tubes
Fig. 4 e Distribution results of the 10 D 10 transparent header.
2202
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
forces to supply the liquid to the top branch (tube). All of these factors affect the distribution. The experiment shows that distribution is usually better for the 10 þ 10 header when the same mass flow rate in one inlet microchannel tube is the same as that in the 5 þ 5 header case. The distribution profiles between the 5 þ 5 (Fig. 3) and the 10 þ 10 headers (Fig. 4) at the same mass flow rate in the header look similar, so the mass flux in the header may be an important parameter in determining the distribution. Therefore, the coefficient of variation of all these distribution results are generalized with liquid mass flux Gl ¼ Gin(1 xin) in the middle of the header, as shown in Fig. 5. The value of s reduces as liquid mass flux increases, i.e. the distribution is better.
Fig. 5 e Coefficient of variation decreases as liquid mass flux increases.
have more liquid when the distribution is more uniform. Regardless, the high quality cases exhibit the worse distribution than the low quality cases. It is just whether the liquid was entrained to branches (tubes) at the higher or at the lower part of the header. In Fig. 4 for the 10 þ 10 header, the trend of the results is very similar as that of the 5 þ 5 header cases: for the same mass flow rate, when the inlet quality increases, the distribution is worse; for the same inlet quality, when the mass flow rate increases, the distribution is usually improved. The mass flow rate in each inlet microchannel tube is a consequence of the design mostly affected by the air side heat transfer calculation. When the number of tubes in each pass is doubled, so is the total inlet mass flow rate. This would help the top tubes to receive more liquid and hence improve the distribution. However, the header is longer than that for the 5 þ 5 configuration. It consequently requires higher inertia
4.2.
Visualization
This section will explain why the distribution profile looks the way it does and how the inlet mass flux, quality and number of tubes affect the distribution based on the visualization. Two flow regimes are identified from the visualization: churn and separated flow. The flow patterns are also listed in Figs. 3 and 4, where “C” denotes the churn flow, while “S” denotes the separated flow. The separated flow is like annular flow, but due to the tubes protrusion, the annulus is not complete. The churn flow typically appears at low qualities; whereas, the separated flow usually occurs at high qualities. Increasing the quality or the mass flux will change the churn flow to separated flow. The flow regimes are shown in Figs. 6 and 7. In the header, the dark part is liquid and the white part is vapor. The black is the aluminum microchannel tubes and epoxy adhesive. Fig. 6 shows the churn flow in the header. Most of the header is the liquid refrigerant with bubbles, but at the top it is almost only vapor. Because the horizontal velocity is small, when the two-phase refrigerant enters into the header, bubbles turn and flow immediately vertically due to the buoyancy
Fig. 6 e Churn flow.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
2203
Fig. 7 e Separated flow.
effects. The liquid falls, forming a liquid pool or interacting with upward flowing vapor. Bubbles stir the liquid and form a local recirculation, although the mean velocity of liquid is upward. It is hard to distinguish the vapor and liquid interface; they are mixed almost homogenously. Such flow regime causes the more uniform distribution at low qualities. Fig. 7 shows the separated flow regime. The void fraction in the header is very high, and thus most volume of the header is taken by the vapor phase, but liquid is present in the form of liquid film along the inner wall. Although the velocity from the inlet microchannel tube is high, it is still easy for vapor to turn and flow upward. However, the liquid jet at the exit from the microchannel tube flows horizontally, hits the wall and forms the liquid film. The high speed vapor and liquid film flow upward along the header. In the top exiting region, the vapor with lighter density is much easier to turn 90 and branch out, but the liquid with larger density and higher momentum tends to run through the header and bypass the first few microchannel tubes. The liquid film is separated from the wall at a certain height. This location h, as shown in Fig. 7, is defined as the liquid separation height. Above the separation height, the flow pattern is locally like churn flow. Some liquid flows horizontally and leaves through the outlet microchannel tubes. Other liquid falls down through the gap between the tube and round header, so that it creates the recirculation in the header. Finally, due to the low momentum, the liquid cannot reach the top and the tubes (branches) there get very little, if any, liquid. The highest liquid level H, as shown in Fig. 7, is defined as the liquid reach. Above the liquid reach, there is almost only vapor. The liquid reach is also observed from churn flow in Fig. 6. Due to such separated flow regime, for example, at high quality and high mass flux, branch 4 has the highest liquid flow rate. The top tube and the bottom tubes have less liquid. It can be seen that the distribution is better in the churn flow. It is because the opportunity of liquid supply to reach
each branch is similar when the liquid occupies most of the header. However, in the separated flow, in only a small part of the header, the liquid is easily available in front of the entrance to the tube. Therefore, generalization of these two flow patterns can be used to evaluate the homogeneity of liquid and vapor phases in the header and may be further the distribution. The generalization is accomplished by considering the two-phase flow in the header as the deviation of two-phase flow in the vertical pipe. It is because: 1) the flow is developing; 2) the mass flux is changing along the header; 3) the tubes protruding in the header obstruct the flow. Thus, the superficial momentum fluxes of liquid and vapor phases rl j2l ¼
½Gin ð1 xin Þ2 rl
Fig. 8 e Flow regime map in the vertical header.
(10)
2204
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Fig. 9 e Local flow regime map in the vertical header.
rv j2v ¼
½Gin xin 2 rv
(11)
as used by Hewitt and Roberts (1969) to generalize the twophase flow in the vertical pipe are applied in attempt to generalize the current results. The flow regime map is shown in Fig. 8 where the transition line is drawn. The local flow regimes between two microchannel tubes are also generalized with this method in Fig. 9. It is noticed
Fig. 11 e Generalization of relative liquid reach with liquid mass flux.
that at high qualities, the local flow regime is separated flow at first. As the two-phase fluid branches out, the superficial vapor momentum is reduced. The local flow regime becomes churn flow. The tubes in the local churn flow region have higher opportunity to receive liquid than those in the separated flow region. This causes maldistribution. However, evaluation of the homogeneity of liquid and vapor phases in the header is not sufficient to evaluate the
Fig. 10 e Liquid reach increases as mass flow rate is higher.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
distribution. The liquid momentum is another important factor. One example is at xin ¼ 0.2 from Fig. 4. At low mass flow rates, although the local flow regime is churn flow throughout the two-phase region in the header, the distribution is very bad since the top tubes have almost no liquid at all. This is due to the liquid reach below the top tube, as shown in Fig. 10. The liquid reach is higher when the inlet mass flow rate increases, thus the distribution is better. The relative liquid reach, H/L, is related to the liquid mass flux Gl in Fig. 11. For the 5 þ 5 header, it is necessary to keep Gl above 20 kg m2 s1, so that the liquid can reach the top tube. Such a condition is Gl > 60 kg m2 s1 for the 10 þ 10 header. This is a necessary but sufficient condition for good distribution. As shown in Fig. 12, the zone where the liquid is present is emphasized with a red square. The liquid reach also changes as the quality increases. Besides, the liquid separation in the separated flow pattern increases when the quality is higher. This is also important for distribution because the higher separation height means the larger region of separated flow and more tubes are bypassed, i.e. the distribution is worse. The relative liquid separation, h/L1, is related to vapor mass flux Gv ¼ Ginxin in Fig. 13. When Gv is below 20 kg m2 s1, the local flow pattern below liquid reach is always churn flow, and
2205
Fig. 13 e Generalization of relative liquid separation with vapor mass flux.
no tubes will be bypassed. This is another necessary but sufficient condition for good distribution. In order to achieve good distribution, both conditions for liquid reach and liquid separation should be satisfied.
Fig. 12 e As quality increases, churn flow becomes separated flow, and liquid separation is higher in separated flow.
2206 4.3.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Distribution correlation
The distribution correlation with liquid take-off ratio G is developed in the two-phase flow region in the header, i.e. below the liquid reach. The notations of the parameters are shown in Fig. 14. Following the ideas of Watanabe et al. (1995) and Byun and Kim (2011), the liquid take-off ratio is related to the vapor phase Reynolds number in the header at immediate upstream, which is the region in the header right before the two-phase fluid exits through the branch tube. The liquid take-off ratio and local vapor phase Reynolds number are calculated based on the corrected experimental data as G¼
_ l;out;i m _ l;M;i m
Rev;M
_ M;i D xM;i m ¼ Amv
(12)
(13)
It is noticed that with this method, the correlation is not a function of local flow regime, but the inlet quality is an important factor (See Fig. 15). The liquid take-off ratio reduces as the Reynolds number increases. It is similar to the findings in Watanabe et al. (1995) and Byun and Kim (2011). Each curvefit correlation shown in Fig. 15 is generated based on the data for each individual quality. They have reasonable accuracy. The R-square value is around 0.8. The new correlation is developed using the least square method. The curve-fit in Fig. 16 contains all the data of both the 5 þ 5 header and the 10 þ 10 header at qualities from 0.2 to 0.95 and mass flux from 21.80 to 127.3 kg m1 s1. It is plotted at different inlet quality condition. It can be seen that the number of tubes in one pass had a slight effect on the liquid
Fig. 14 e Header schematic and notations.
Fig. 15 e Liquid take-off ratio G as a function of upstream Reynolds number Re and inlet quality x. Each line (correlation) is derived from data in defined range of quality.
Fig. 16 e Results of the correlation presented by equation (14) plotted with quality as a parameter. a) Curve-fit, b) Accuracy.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
take-off ratio. Thus, the correlation for R410A, D ¼ 15 mm with a 5 or 10 inlet and a 5 or 10 outlet half depth protruded microchannel tubes at qualities from 0.2 to 0.95 and mass flux from 21.80 to 127.3 kg m1 s1 is 1:447 Rev;M G ¼ 421989x1:248 in
(14)
The liquid take-off ratio can be incorporated into the heat exchanger model in Tuo et al. (2012) to evaluate heat exchanger performance. Although the accuracy of the generalization may not be very good, it is the best method available from the literature. In future work, the authors should consider investigating a more general way in which they would consider the local flow regimes as presented in this paper.
5.
Conclusion
This paper presented R410A distribution in a typical microchannel heat exchanger geometry with a vertical header. The distribution usually gets worse as the quality increases. At low qualities (e.g. xin 0.6), the distribution usually improves as the mass flow rate gets higher. At high qualities (e.g. xin 0.8), the distribution is better at some intermediate mass flow rate, but generally speaking the distributions are poor. Due to the increase in mass flux, a great number of the microchannel tubes improve the distribution although a longer header requires higher momentum for the liquid to reach the top tube. Generalization of s shows that as the liquid mass flux increases, the distribution gets better. The churn and separated flow in the header are identified. The separated flow in the header is found as the inlet quality and/or mass flux increases. The local churn flow throughout the header generates better distribution because the vapor and liquid phases are more homogeneous. These two flow patterns are generalized using superficial liquid and vapor momentum fluxes. Based on the visualization, the axial momentum is another important factor that affects the refrigerant distribution. The highest liquid level, defined as liquid reach, is generalized with the liquid mass flux. In order for the liquid to reach the top tube, Gl must be greater than 20 kg m2 s1 for the 5 þ 5 header and Gl must be greater than 60 kg m2 s1 for the 10 þ 10 header. In the separated flow, the liquid separation is generalized with the vapor mass flux. To avoid the liquid bypassing the first outlet tube, Gv must be less than 20 kg m2 s1. An empirical liquid distribution correlation was obtained using the liquid take-off ratio. It can be incorporated into existing microchannel heat exchanger model. The transparent headers provide the insight in the flow pattern in the header and allow for building the relationship between the flow regime (visual) and distribution (measured by mass flow rate). Even no direct correlation between them is established at this point. Very useful insights and indications were generated. We hope it will help us in future developments of understanding the adiabatic two-phase flow in the header and heat exchanger.
2207
Acknowledgment This project was completed in 2010. It was conducted at the Air Conditioning and Refrigeration Center at the University of Illinois at Urbana-Champaign, sponsored by Daikin Industries, Ltd., and also supported by Creative Thermal Solutions. Furthermore, the authors would like to thank Dr. Hyunyong Kim from Daikin Industries, Ltd. for his help and advice on this project.
references
Bowers, C.D., Newell, T.A., Hrnjak, P.S., 2006. Experimental Investigation of Two-phase Refrigerant Distribution in a Microchannel Manifold. Technical Report TR-245. University of Illinois at Urbana-Champaign, Urbana, USA.. Brix, W., Kœrn, M.R., Elmegaard, B., 2009. Modeling refrigerant distribution in microchannel evaporators. Int. J. Refrigeration 32, 1736e1743. Brix, W., Kœrn, M.R., Elmegaard, B., 2010. Modeling distribution of evaporating CO2 in parallel Minichannels. Int. J. Refrigeration 33, 1086e1094. Byun, H.W., Kim, N.H., 2011. Refrigerant distribution in a parallel flow heat exchanger having vertical headers and heated horizontal tubes. Exp. Therm. Fluid Sci. 35, 920e930. Cho, H., Cho, K., 2004. Mass flow rate distribution and phase separation of R-22 in multi-microchannel tubes under adiabatic condition. Microscale Thermophysical Eng. 8 (2), 129e139. Dschida, M., Hrnjak, P.S., 2008. Evaluation of Microchannel Heat Exchangers for a Residential Mini-split Type Air-Conditioning/ Heat Pump System. Contract Report CR-67. University of Illinois at Urbana-Champaign, Urbana, USA. EES, 2010. Engineering Equation Solver. Academic Professional Version. F-Chart Software, Middleton, WI, USA. Fei, P., Hrnjak, P.S., 2004. Adiabatic Developing Two-phase Refrigerant Flow in Manifolds of Heat Exchangers. Technical Report TR-225. University of Illinois at Urbana-Champaign, Urbana, USA. http://hdl.handle.net/2142/12316. Hewitt, G.F., Roberts, D.N., 1969. Studies of Two-phase Flow Patterns by Simultaneous X-ray and Flash Photography. UKAEA Report AERE-M 2159. HMSO. Hrnjak, P., 2004. Developing adiabatic two phase flow in headers e distribution issue in parallel flow microchannel heat exchangers. Heat Transfer Eng. 25 (3), 61e68. Hwang, Y., Jin, D.H., Radermacher, R., 2007. Refrigerant distribution in minichannel evaporator manifolds. HVAC&R Res. 13 (4), 543e555. Kulkarni, T., Bullard, C.W., Cho, K., 2004. Header design tradeoffs in microchannel evaporators. Appl. Therm. Eng. 24, 759e776. Lee, J.K., 2009. Two-phase flow behavior inside a header connected to multiple parallel channels. Exp. Therm. Fluid Sci. 33, 195e202. Moura, F.L., 1990. Experimental and numerical study on the twophase flow distribution between two passes of a heat exchanger. ASME Fluids Eng. Division (Publication) 99, 267e273. Song, S., Bullard, C.W., Hrnjak, P.S., 2002. Frost deposition and refrigerant distribution in microchannel heat exchangers. ASHRAE Trans. 108 (2), 944e953. Tompkins, D.M., Newell, T.A., Hrnjak, P.S., 2002. Single Phase, Two-phase Modeling; X-ray Visualization for a Microchannel Manifold Distribution System. Technical Report TR-206. University of Illinois at Urbana-Champaign, Urbana, USA. .
2208
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Tuo, H., Bielskus, A., Hrnjak, P.S., 2012. Experimentally Validated Model of Refrigerant Distribution in a Parallel Microchannel Evaporator. SAE World Congress, Detroit, MI.. Paper 2012-010321. Vist, S., Pettersen, J., 2004. Two-phase flow distribution in compact heat exchanger manifolds. Exp. Therm. Fluid Sci. 28, 209e215.
Watanabe, M., Katsuta, M., Nagata, K., 1995. General characteristics of two-phase flow distribution in a multipass tube. Heat Transfer-Japanese Res. 24, 32e44. Webb, R.L., Chung, K., 2005. Two-phase flow distribution to tubes of parallel flow air-cooled heat exchangers. Heat Transf. Eng. 26 (4), 3e18.
Available online at www.sciencedirect.com
w w w . i i fi i r . o r g
journal homepage: www.elsevier.com/locate/ijrefrig
Refrigerant distribution in the vertical header of the microchannel heat exchanger e Measurement and visualization of R410A flow Yang Zou a, Pega S. Hrnjak a,b,* a
Air Conditioning and Refrigeration Center, Department of Mechanical Science and Engineering, University of Illinois at Urbana Champaign, 1206 West Green Street, Urbana, IL 61801, United States b Creative Thermal Solutions Inc., 2209 North Willow Road, Urbana, IL 61802, United States
article info
abstract
Article history:
This paper presents the R410A adiabatic upward flow in three vertical headers of micro-
Received 26 November 2012
channel heat exchangers. All microchannel tubes are inserted into the half depth. The ob-
Received in revised form
jectives are to explore what affects R410A distribution and attempt to predict the distribution.
30 March 2013
R410A is circulated into the header through the (5 or 10) tubes in the bottom pass and exits
Accepted 30 April 2013
through the (5 or 10) tubes in the top pass. It represents the flow in the outdoor heat exchanger
Available online 9 May 2013
(usually used as the condenser) when it is used as the evaporator in the heat pump mode of reversible systems. The quality was typically varied from 0.2 to 0.8 and the mass flow rate
Keywords:
from 1.5 to 4.5 kg h1 per tube. The best distribution was observed at high mass flux and low
Two-phase flow
quality. The experiment and visualization reveals the flow pattern effects in terms of ho-
Refrigerant distribution
mogeneity and momentum. The churn flow had better distribution since the two-phase
Vertical header
mixture was more homogenous and the distribution was better at high mass flux in the
Microchannel
header because the higher momentum liquid was able to reach the top exit tube.
Heat exchanger
ª 2013 Elsevier Ltd and IIR. All rights reserved.
Distribution du frigorige`ne dans le collecteur vertical d’un e´changeur de chaleur a` microcanaux e Mesures et visualisation de l’e´coulement du R410A Mots cle´s : e´coulement diphasique ; distribution du frigorige`ne ; collecteur vertical ; microcanal ; e´changeur de chaleur
1.
Introduction
The microchannel heat exchangers have come to the frontier of stationary air conditioning and refrigeration applications
after its success in the mobile air conditioning systems, for their advantages in compactness and higher air-side heat transfer coefficients. In the reversible systems, the microchannel heat exchanger functions as both the evaporator and
* Corresponding author. Air Conditioning and Refrigeration Center, Department of Mechanical Science and Engineering, University of Illinois at Urbana Champaign, 1206 West Green Street, Urbana, IL 61801, United States. Tel.: þ1 217 244 6377; fax: þ1 217 244 6534. E-mail address: [email protected] (P.S. Hrnjak). 0140-7007/$ e see front matter ª 2013 Elsevier Ltd and IIR. All rights reserved. http://dx.doi.org/10.1016/j.ijrefrig.2013.04.021
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Nomenclature A D CF G H h i j L LF m n Q P Re T U x y
2
Cross section area in the header (m ) Internal diameter of the header (m) Correction factor () Mass flux (kg m2 s1) Liquid reach (m) Liquid separation height (m) Enthalpy (kJ kg1) Superficial velocity (m s1) Length of the top exit part in header (m) Liquid fraction () Mass flow rate (g s-1) number of the outlet tubes () Power of heaters (kW) Pressure (kPa) Reynolds number () Temperature (K) A function Quality () Parameter
condenser. The outdoor heat exchanger, typically a multipass microchannel heat exchanger, has vertical headers to reduce the cost and allow for easier bending when a “U-shaped” condenser is needed. In the cooling mode, the outdoor heat exchanger functions as the condenser and the overall flow is downward. In the heating mode, it functions as the evaporator and the refrigerant flows in the reverse direction. Refrigerant maldistribution due to the two-phase flow in the header deteriorates the heat exchanger’s performance, and consequently reduces the system efficiency. Kulkarni et al. (2004) simulated the effect of R410A maldistribution induced by the pressure drop in the horizontal header. The microchannel evaporator’s capacity was reduced by 20%. Brix et al. (2009, 2010) modeled R134a and R744 distribution in two microchannel tubes, respectively. Due to maldistribution, the capacity was reduced by 23% and 18%, respectively. Good designs that provide reasonable distribution in either the air conditioning or the heat pump modes are still a challenge, although it has been extensively studied. Hrnjak (2004) and Webb and Chung (2005) have reviewed the refrigerant distribution issues in the header. Combined with others’ findings, it can be conceived that maldistribution is a very complex problem that is affected by numerous parameters: header geometry and orientation, fluid properties and inlet conditions, etc. Specifically, in their studies of horizontal headers, Fei and Hrnjak (2004), Vist and Pettersen (2004), Bowers et al. (2006) and Hwang et al. (2007) showed that the flow regimes in the header, which is affected by several conditions, had a strong influence on liquid distribution into the branch tubes in two-phase flow. In the studies of vertical headers, Moura (1990) examined airewater distribution in the second-pass header. In either the upward or downward flow in the header, phase separation occurred. The flow patterns strongly affected the distribution. Watanabe et al. (1995) investigated the distribution of R11 in a header with five round branch tubes. Along the upward flow, the flow regime transited from annular to slug, or froth flow, in the header.
G m r s
2197
Liquid take-off ratio () Viscosity (kg m1 s1) Density (kg m3) Coefficient of variation ()
Superscripts c Corrected meas Measured Subscripts ave Average pressure i Branch number in In the middle of the header and at the smallest area of the last inlet tube l Liquid M Main pipe (header) out Out from the header sup Superheated point sub Subcooled point v vapor
Increasing the quality to 0.3 improved the distribution because the higher velocity enabled liquid to reach the top tube. Song et al. (2002) manifested CO2 maldistribution in a multi-pass outdoor microchannel coil during the heating mode through frosting patterns. The balance between inertial, gravitational and shear forces generated two different distributions: 1) the liquid was only present in the bottom tubes at low qualities; 2) the top and bottom tubes received less liquid than the middle tubes at high qualities. Cho and Cho (2004) tested R22 distribution in the header with the microchannel tubes. Most liquid was in the bottom due to gravity regardless of the inlet types and inlet quality. Through infrared images, Dschida and Hrnjak (2008) showed R410A maldistribution in a multi-pass microchannel heat exchanger during the heat pump mode. The superheated zone always appeared at the top in each pass, i.e. less liquid at the top, if any. The visualization of upward air and water flow in the header from Lee (2009) showed that three distinct regions were formed with different flow patterns, which caused three different distribution patterns. Byun and Kim (2011) tested and visualized R410A distribution in both the inlet and second-pass headers at one inlet quality condition. For the inlet header, a liquid pool was formed at the bottom. For the second pass header, an upward-flow liquid film was formed in the header. Because of the high axial momentum, most liquid was taken by the middle tubes. Increasing the mass flux resulted in most liquid in the top tubes. Although this study is similar to Byun and Kim (2011), this study uses a different experiment method so that the mass flow rate in each tube is measured and this study covers a much wider range of inlet mass flux and quality. The effects of tubes number is also explored. To predict the refrigerant distribution, either numerical simulation or empirical correlation was developed. Moura (1990) and Tompkins et al. (2002) developed their models based on finite volume method. Fei and Hrnjak (2004) conducted CFD simulation with the commercial software FLUENT 6. Lee (2009) considered the vertical header as a series of T-
2198
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
junctions, and predicted the distribution based on the studies of two-phase flow at T-junction. Watanabe et al. (1995) and Byun and Kim (2011) developed empirical correlations based on the defined liquid and vapor take-off ratio. It is the ratio of liquid (or vapor) mass flow rate in the branch tube to the liquid (or vapor) mass flow rate in the main pipe (header) at immediate upstream, which is the region in the header right before the two-phase fluid exits through the branch tube. The liquid (or vapor) take-off ratio is related to the vapor phase Reynolds number in the main pipe at immediate upstream. This method will be used to develop an empirical correlation in this paper. It is noteworthy that most of the previous tests were done in the horizontal headers and low qualities (i.e. <0.5) because the indoor heat exchanger and/or inlet header were usually studied. However, typical outdoor microchannel heat exchangers have vertical headers and at least two passes. The two-phase refrigerant is redistributed in the intermediate header in which case the distribution phenomenon is more complex but important. The previous studies on multi-pass heat exchangers with nonintrusive methods, such as infrared imaging and frost accumulation, did not provide enough information to quantify refrigerant distribution. Thus, in this paper, R410A distribution in the vertical intermediate header of a multi-pass heat exchanger (with capacity of 0.5e2.5 kW) was studied by adding individual flow rate measurement and visualization.
2.
Experimental method
The test loop was constructed to study R410A distribution in the microchannel heat exchanger, as shown in Fig. 1. The liquid refrigerant was pumped into the inlet header while the inlet mass flow rate min was measured by the mass flow meter (Micromotion D40, 0.75%). The mass flow rate was controlled by a bypass valve and variable frequency drive. The temperature Tsub (immersed thermocouple, 0.5 C) and pressure Psub (Sensotec TJE, 1% FS) at the inlet to the first header were
measured to determine the subcooling. The subcooled liquid was assumed to distribute evenly into the microchannel tubes in the bottom pass, where the refrigerant was heated to the desired quality while the heaters are insulated. The set of six heaters per microchannel tube were providing uniform heat supply. The power needed to determine the inlet quality was measured by the watt transducer (Ohio Semitronics GW5024CX5, 0.2% FS). After the refrigerant entered into the lower part of the header, it turned 90 to flow upward and reach the upper part of the header. Due to maldistribution, different amounts of liquid exited through the microchannel tubes in the top pass. In these tubes, the refrigerant was heated again to provide equal superheat at the exit. Each tube was heated by six heaters and insulated. Each set of six heaters was individually controlled to provide adequate power to generate equal superheat, and was also individually measured by watt transducers (Ohio Semitronics GW5024CX5, 0.2% FS) to determine inlet quality into each exit tube. The temperature Tsup,i (immersed thermocouple, 0.5 C) in each tube and pressure Psup (Sensotec TJE, 1% FS) were measured to determine the superheat. The mass flow rate of the single-phase superheated vapor was measured individually as the total mass flow rate for each microchannel tube by the mass flow meter (Micromotion D06, 0.15%). The vapor was then brought to the condenser. With the help of the receiver and the subcooler, the subcooled liquid was provided to the pump. The thermal physical and thermodynamic properties are calculated in EES (2010). Enthalpy at the superheated point isup,i was determined by Tsup,i and Psup. The outlet enthalpy from the header, i.e. inlet to each exit tube, iout,i, was determined as iout;i ¼ isup;i
Q_ out;i _ out;i m
(1)
The pressure in the header Pave is estimated as the average of the subcooled and superheating pressures. The outlet quality from the header xout,i can be found from iout,i and Pave. Then, the liquid mass flow rate in each tube can be obtained as
Fig. 1 e System schematics.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
_ out;i 1 xout;i _ l;out;i ¼ m m
(2)
A high speed camera, Phantom v4.2, is used for visualizing the flow in the transparent header. The exposure time of the camera was set from 80 ms to 100 ms. The framing rate was at 2200 frames per second. The resolution was set as 512 512 or 256 512 pixels. Three circular headers were examined, each with all the microchannel tubes inserted to the half depth: 1) the aluminum header with five inlet and five exit microchannel tubes as used in industry (5 þ 5 aluminum header), 2) the transparent replica of the first (5 þ 5 transparent header), and 3) the transparent header with ten inlet and ten exit microchannel tubes (10 þ 10 transparent header) to study the effect of increasing microchannel number and header length. The transparent headers were made of PVC tube. The gap between the microchannel tube and PVC tube was sealed by a special epoxy adhesive. It is the black material on the right part of the header in the images. In addition, an epoxy block was made to increase the pressure tolerance while still ensuring the transparency outside the PVC tube. The measured inner diameter of the aluminum header is 14.94 mm, and that of the transparent header is 15.44 mm. The 5 þ 5 header is 170 mm long, and the 10 þ 10 header is 300 mm long. The tube pitch is 13 mm. The tubes are 1200 mm long. As shown in Fig. 1, each microchannel tube has 17 rectangular channels. Based on the manufacturer data, each microchannel is 0.5 mm 0.54 mm. The hydraulic diameter is calculated to be about 0.5 mm. To examine the case of ten inlet and ten exit tubes using the same header diameter (different circuiting), modification of the facility was made with intention to maintain the same number of mass flow meters (five) and the same heaters and control strategy. Two neighboring exit tubes were combined into one with a “Y” connection and the refrigerant was heated together to be superheated. Assuming that this change affected resolution of the distribution parameters only, the essence of the results should not change. Therefore, only five liquid fraction results were shown, whereas actually each one represented the liquid amount in the neighboring two branches. As shown in Fig. 1, in order to facilitate the connection of 10 þ 10 header, only five inlet microchannel tubes were connected and allowed the refrigerant to pass through implicitly assuming the insignificant effect of inlet difference. The other five inlet microchannel tubes were blocked and just served as obstructions in the header. The averaged saturation temperature was about 5 C in the header. The inlet quality ranged from 0.2 to 0.8. The results of xin ¼ 0 and 1 were taken to provide two asymptotes, and those of xin ¼ 0.95 were to offer more typical situations at high quality. The mass flow rate was from 1.5 to 4.5 kg h1 for each microchannel tube. Thus, the inlet mass flow rate ranged from 2.14 to 6.25 g s1 for the 5 þ 5 header and from 4.28 to 12.50 g s1 for the 10 þ 10 header. Tests at 7.27 g s1 for the 5 þ 5 header were taken in attempt to reach better distribution. Tests at 2.22 g s1 for the 10 þ 10 header were taken to compare the results with those at 2.14 g s1 for the 5 þ 5 header. From Bowers et al. (2006), the mass flux defined by the smallest cross sectional area in the header is more representative when the protrusion is present, so the maximum mass
2199
flux in the header Gin, as shown in Fig. 1, will be used in the following analysis. And Gin ranged from 21.80 to 63.67 kg m2 s1 for the 5 þ 5 headers and from 22.62 to 127.3 kg m1 s1 for the 10 þ 10 header.
3.
Data reduction
Two metrics are used to evaluate the distribution. The first one is coefficient of variation s. It is the dimensionless standard deviation, which is defined as s¼
1 ml
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi n 2 1X _ l;out;i m l m n 1
(3)
The uniform distribution is described by s ¼ 0. The worst distribution, for this case (with five results), is when s ¼ 2. The advantage of this metric is that one number is able to characterize the goodness of distribution. However, in order to illustrate the distribution profile, it is necessary to apply the liquid fraction, LFi ¼
_ l;out;i m n P _ l;out;i m
(4)
i
For the case with five results, the uniform distribution is described by LFi ¼ 0.2 in every tube. The uncertainty propagation analysis of s and LFi was carried out in the EES (2010). It was based on vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u N uX vU dyi where U ¼ U y1 ; y2 ; y3 ; ::: dU ¼ t vy i i¼1
(5)
Thus, the uncertainty of s was within 4.5%. The uncertainty of liquid fraction is usually below 5%. Due to the uncertainty of the instruments such as the mass flow meter and watt transducer, the sum of the mass flow rates in the outlet tubes is not the same as the inlet mass flow rate. To analyze the data and develop a correlation, the mass flow rate in the outlet tube is corrected as in equations (6) and (7). _ meas _ cout;i ¼ m m out;i CF
(6)
_ meas m CF ¼ Pn in meas _ i¼1 mout;i
(7)
In addition, the liquid mass flow rates in the outlet tubes are adjusted such that the sum is equal to the inlet liquid mass flow rate, as in equations (8) and (9). _ meas _ cl;out;i ¼ m m l;out;i CFl
CFl ¼
_ meas 1 xmeas m in in Pn _ meas i¼1 ml;out;i
(8)
(9)
The typical CF value is around 0.9. For the 10 þ 10 header, only five liquid flow rate results were obtained. To estimate the flow rate in each tube for further analysis, it is assumed that the mass flow rate (either liquid or vapor) of the neighboring two tubes is equal.
2200
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
4.
Results and discussion
4.1.
Distribution profiles
Figs. 2e4 show the distribution profiles and the coefficient of variation of the 5 þ 5 aluminum header (Fig. 2), 5 þ 5 transparent header (Fig. 3) and 10 þ 10 transparent header (Fig. 4). The darkness of the bar color represents different branches (tubes), the pale being the lowest exit branch and the dark being the highest exit branch. The best distribution indicated
by s is at xin ¼ 0.2 (the lowest) and the highest inlet mass flux among the test conditions for each header. The results of the 5 þ 5 aluminum and transparent headers are very similar (Figs. 2 and 3). The small discrepancy may be due to the different internal areas. In Figs. 2(a) and 3, the liquid fraction of the top branch is higher as the mass flow rate increases and at xin ¼ 0.6, 0.8 and 0.95, the liquid fractions of the bottom branches reduce at the same time, whereas at xin ¼ 0.2 and 0.4, the liquid fractions of the bottom branches do not change much. On the other hand, the liquid fraction of the highest branch reduces as the quality increases. Meanwhile,
Fig. 2 e Distribution results of the 5 D 5 aluminum header. a) Liquid fraction, b) Coefficient of variation.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
2201
Fig. 3 e Distribution results of the 5 D 5 transparent header.
for the other branches, as quality increases, 1) at min ¼ 2.14 and 3.17 g s1, the bottom branches have higher and higher liquid fractions; 2) at min ¼ 4.19 g s1, more liquid leaves from the middle tubes; 3) at min ¼ 5.22, 6.25 and 7.27 g s1, the liquid fraction of branch 4 is highest while the others have less and less. The values of the coefficient of variation in Fig. 2(b) show that: 1) at the fixed mass flow rate, the distribution usually
deteriorates as quality increases; 2) at low qualities, the distribution improves with increasing mass flux; 3) at high qualities, the distribution is better at some intermediate mass flow rate. It gets worse as min either increases or decreases. It is because as min increases, the bottom tubes liquid rich transited to the top tubes liquid rich. During this transition, there is a case that neither the bottom tubes nor the top tubes
Fig. 4 e Distribution results of the 10 D 10 transparent header.
2202
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
forces to supply the liquid to the top branch (tube). All of these factors affect the distribution. The experiment shows that distribution is usually better for the 10 þ 10 header when the same mass flow rate in one inlet microchannel tube is the same as that in the 5 þ 5 header case. The distribution profiles between the 5 þ 5 (Fig. 3) and the 10 þ 10 headers (Fig. 4) at the same mass flow rate in the header look similar, so the mass flux in the header may be an important parameter in determining the distribution. Therefore, the coefficient of variation of all these distribution results are generalized with liquid mass flux Gl ¼ Gin(1 xin) in the middle of the header, as shown in Fig. 5. The value of s reduces as liquid mass flux increases, i.e. the distribution is better.
Fig. 5 e Coefficient of variation decreases as liquid mass flux increases.
have more liquid when the distribution is more uniform. Regardless, the high quality cases exhibit the worse distribution than the low quality cases. It is just whether the liquid was entrained to branches (tubes) at the higher or at the lower part of the header. In Fig. 4 for the 10 þ 10 header, the trend of the results is very similar as that of the 5 þ 5 header cases: for the same mass flow rate, when the inlet quality increases, the distribution is worse; for the same inlet quality, when the mass flow rate increases, the distribution is usually improved. The mass flow rate in each inlet microchannel tube is a consequence of the design mostly affected by the air side heat transfer calculation. When the number of tubes in each pass is doubled, so is the total inlet mass flow rate. This would help the top tubes to receive more liquid and hence improve the distribution. However, the header is longer than that for the 5 þ 5 configuration. It consequently requires higher inertia
4.2.
Visualization
This section will explain why the distribution profile looks the way it does and how the inlet mass flux, quality and number of tubes affect the distribution based on the visualization. Two flow regimes are identified from the visualization: churn and separated flow. The flow patterns are also listed in Figs. 3 and 4, where “C” denotes the churn flow, while “S” denotes the separated flow. The separated flow is like annular flow, but due to the tubes protrusion, the annulus is not complete. The churn flow typically appears at low qualities; whereas, the separated flow usually occurs at high qualities. Increasing the quality or the mass flux will change the churn flow to separated flow. The flow regimes are shown in Figs. 6 and 7. In the header, the dark part is liquid and the white part is vapor. The black is the aluminum microchannel tubes and epoxy adhesive. Fig. 6 shows the churn flow in the header. Most of the header is the liquid refrigerant with bubbles, but at the top it is almost only vapor. Because the horizontal velocity is small, when the two-phase refrigerant enters into the header, bubbles turn and flow immediately vertically due to the buoyancy
Fig. 6 e Churn flow.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
2203
Fig. 7 e Separated flow.
effects. The liquid falls, forming a liquid pool or interacting with upward flowing vapor. Bubbles stir the liquid and form a local recirculation, although the mean velocity of liquid is upward. It is hard to distinguish the vapor and liquid interface; they are mixed almost homogenously. Such flow regime causes the more uniform distribution at low qualities. Fig. 7 shows the separated flow regime. The void fraction in the header is very high, and thus most volume of the header is taken by the vapor phase, but liquid is present in the form of liquid film along the inner wall. Although the velocity from the inlet microchannel tube is high, it is still easy for vapor to turn and flow upward. However, the liquid jet at the exit from the microchannel tube flows horizontally, hits the wall and forms the liquid film. The high speed vapor and liquid film flow upward along the header. In the top exiting region, the vapor with lighter density is much easier to turn 90 and branch out, but the liquid with larger density and higher momentum tends to run through the header and bypass the first few microchannel tubes. The liquid film is separated from the wall at a certain height. This location h, as shown in Fig. 7, is defined as the liquid separation height. Above the separation height, the flow pattern is locally like churn flow. Some liquid flows horizontally and leaves through the outlet microchannel tubes. Other liquid falls down through the gap between the tube and round header, so that it creates the recirculation in the header. Finally, due to the low momentum, the liquid cannot reach the top and the tubes (branches) there get very little, if any, liquid. The highest liquid level H, as shown in Fig. 7, is defined as the liquid reach. Above the liquid reach, there is almost only vapor. The liquid reach is also observed from churn flow in Fig. 6. Due to such separated flow regime, for example, at high quality and high mass flux, branch 4 has the highest liquid flow rate. The top tube and the bottom tubes have less liquid. It can be seen that the distribution is better in the churn flow. It is because the opportunity of liquid supply to reach
each branch is similar when the liquid occupies most of the header. However, in the separated flow, in only a small part of the header, the liquid is easily available in front of the entrance to the tube. Therefore, generalization of these two flow patterns can be used to evaluate the homogeneity of liquid and vapor phases in the header and may be further the distribution. The generalization is accomplished by considering the two-phase flow in the header as the deviation of two-phase flow in the vertical pipe. It is because: 1) the flow is developing; 2) the mass flux is changing along the header; 3) the tubes protruding in the header obstruct the flow. Thus, the superficial momentum fluxes of liquid and vapor phases rl j2l ¼
½Gin ð1 xin Þ2 rl
Fig. 8 e Flow regime map in the vertical header.
(10)
2204
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Fig. 9 e Local flow regime map in the vertical header.
rv j2v ¼
½Gin xin 2 rv
(11)
as used by Hewitt and Roberts (1969) to generalize the twophase flow in the vertical pipe are applied in attempt to generalize the current results. The flow regime map is shown in Fig. 8 where the transition line is drawn. The local flow regimes between two microchannel tubes are also generalized with this method in Fig. 9. It is noticed
Fig. 11 e Generalization of relative liquid reach with liquid mass flux.
that at high qualities, the local flow regime is separated flow at first. As the two-phase fluid branches out, the superficial vapor momentum is reduced. The local flow regime becomes churn flow. The tubes in the local churn flow region have higher opportunity to receive liquid than those in the separated flow region. This causes maldistribution. However, evaluation of the homogeneity of liquid and vapor phases in the header is not sufficient to evaluate the
Fig. 10 e Liquid reach increases as mass flow rate is higher.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
distribution. The liquid momentum is another important factor. One example is at xin ¼ 0.2 from Fig. 4. At low mass flow rates, although the local flow regime is churn flow throughout the two-phase region in the header, the distribution is very bad since the top tubes have almost no liquid at all. This is due to the liquid reach below the top tube, as shown in Fig. 10. The liquid reach is higher when the inlet mass flow rate increases, thus the distribution is better. The relative liquid reach, H/L, is related to the liquid mass flux Gl in Fig. 11. For the 5 þ 5 header, it is necessary to keep Gl above 20 kg m2 s1, so that the liquid can reach the top tube. Such a condition is Gl > 60 kg m2 s1 for the 10 þ 10 header. This is a necessary but sufficient condition for good distribution. As shown in Fig. 12, the zone where the liquid is present is emphasized with a red square. The liquid reach also changes as the quality increases. Besides, the liquid separation in the separated flow pattern increases when the quality is higher. This is also important for distribution because the higher separation height means the larger region of separated flow and more tubes are bypassed, i.e. the distribution is worse. The relative liquid separation, h/L1, is related to vapor mass flux Gv ¼ Ginxin in Fig. 13. When Gv is below 20 kg m2 s1, the local flow pattern below liquid reach is always churn flow, and
2205
Fig. 13 e Generalization of relative liquid separation with vapor mass flux.
no tubes will be bypassed. This is another necessary but sufficient condition for good distribution. In order to achieve good distribution, both conditions for liquid reach and liquid separation should be satisfied.
Fig. 12 e As quality increases, churn flow becomes separated flow, and liquid separation is higher in separated flow.
2206 4.3.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Distribution correlation
The distribution correlation with liquid take-off ratio G is developed in the two-phase flow region in the header, i.e. below the liquid reach. The notations of the parameters are shown in Fig. 14. Following the ideas of Watanabe et al. (1995) and Byun and Kim (2011), the liquid take-off ratio is related to the vapor phase Reynolds number in the header at immediate upstream, which is the region in the header right before the two-phase fluid exits through the branch tube. The liquid take-off ratio and local vapor phase Reynolds number are calculated based on the corrected experimental data as G¼
_ l;out;i m _ l;M;i m
Rev;M
_ M;i D xM;i m ¼ Amv
(12)
(13)
It is noticed that with this method, the correlation is not a function of local flow regime, but the inlet quality is an important factor (See Fig. 15). The liquid take-off ratio reduces as the Reynolds number increases. It is similar to the findings in Watanabe et al. (1995) and Byun and Kim (2011). Each curvefit correlation shown in Fig. 15 is generated based on the data for each individual quality. They have reasonable accuracy. The R-square value is around 0.8. The new correlation is developed using the least square method. The curve-fit in Fig. 16 contains all the data of both the 5 þ 5 header and the 10 þ 10 header at qualities from 0.2 to 0.95 and mass flux from 21.80 to 127.3 kg m1 s1. It is plotted at different inlet quality condition. It can be seen that the number of tubes in one pass had a slight effect on the liquid
Fig. 14 e Header schematic and notations.
Fig. 15 e Liquid take-off ratio G as a function of upstream Reynolds number Re and inlet quality x. Each line (correlation) is derived from data in defined range of quality.
Fig. 16 e Results of the correlation presented by equation (14) plotted with quality as a parameter. a) Curve-fit, b) Accuracy.
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
take-off ratio. Thus, the correlation for R410A, D ¼ 15 mm with a 5 or 10 inlet and a 5 or 10 outlet half depth protruded microchannel tubes at qualities from 0.2 to 0.95 and mass flux from 21.80 to 127.3 kg m1 s1 is 1:447 Rev;M G ¼ 421989x1:248 in
(14)
The liquid take-off ratio can be incorporated into the heat exchanger model in Tuo et al. (2012) to evaluate heat exchanger performance. Although the accuracy of the generalization may not be very good, it is the best method available from the literature. In future work, the authors should consider investigating a more general way in which they would consider the local flow regimes as presented in this paper.
5.
Conclusion
This paper presented R410A distribution in a typical microchannel heat exchanger geometry with a vertical header. The distribution usually gets worse as the quality increases. At low qualities (e.g. xin 0.6), the distribution usually improves as the mass flow rate gets higher. At high qualities (e.g. xin 0.8), the distribution is better at some intermediate mass flow rate, but generally speaking the distributions are poor. Due to the increase in mass flux, a great number of the microchannel tubes improve the distribution although a longer header requires higher momentum for the liquid to reach the top tube. Generalization of s shows that as the liquid mass flux increases, the distribution gets better. The churn and separated flow in the header are identified. The separated flow in the header is found as the inlet quality and/or mass flux increases. The local churn flow throughout the header generates better distribution because the vapor and liquid phases are more homogeneous. These two flow patterns are generalized using superficial liquid and vapor momentum fluxes. Based on the visualization, the axial momentum is another important factor that affects the refrigerant distribution. The highest liquid level, defined as liquid reach, is generalized with the liquid mass flux. In order for the liquid to reach the top tube, Gl must be greater than 20 kg m2 s1 for the 5 þ 5 header and Gl must be greater than 60 kg m2 s1 for the 10 þ 10 header. In the separated flow, the liquid separation is generalized with the vapor mass flux. To avoid the liquid bypassing the first outlet tube, Gv must be less than 20 kg m2 s1. An empirical liquid distribution correlation was obtained using the liquid take-off ratio. It can be incorporated into existing microchannel heat exchanger model. The transparent headers provide the insight in the flow pattern in the header and allow for building the relationship between the flow regime (visual) and distribution (measured by mass flow rate). Even no direct correlation between them is established at this point. Very useful insights and indications were generated. We hope it will help us in future developments of understanding the adiabatic two-phase flow in the header and heat exchanger.
2207
Acknowledgment This project was completed in 2010. It was conducted at the Air Conditioning and Refrigeration Center at the University of Illinois at Urbana-Champaign, sponsored by Daikin Industries, Ltd., and also supported by Creative Thermal Solutions. Furthermore, the authors would like to thank Dr. Hyunyong Kim from Daikin Industries, Ltd. for his help and advice on this project.
references
Bowers, C.D., Newell, T.A., Hrnjak, P.S., 2006. Experimental Investigation of Two-phase Refrigerant Distribution in a Microchannel Manifold. Technical Report TR-245. University of Illinois at Urbana-Champaign, Urbana, USA.
2208
i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 3 6 ( 2 0 1 3 ) 2 1 9 6 e2 2 0 8
Tuo, H., Bielskus, A., Hrnjak, P.S., 2012. Experimentally Validated Model of Refrigerant Distribution in a Parallel Microchannel Evaporator. SAE World Congress, Detroit, MI.. Paper 2012-010321. Vist, S., Pettersen, J., 2004. Two-phase flow distribution in compact heat exchanger manifolds. Exp. Therm. Fluid Sci. 28, 209e215.
Watanabe, M., Katsuta, M., Nagata, K., 1995. General characteristics of two-phase flow distribution in a multipass tube. Heat Transfer-Japanese Res. 24, 32e44. Webb, R.L., Chung, K., 2005. Two-phase flow distribution to tubes of parallel flow air-cooled heat exchangers. Heat Transf. Eng. 26 (4), 3e18.